CobSeg: Coherence Boundary Modeling for Dialogue Topic Segmentation
This work provides an incremental improvement in dialogue topic segmentation for human-AI collaborative applications by better identifying heterogeneous boundary cues.
This paper addresses dialogue topic segmentation by proposing CobSeg, a multi-branch architecture that models coherence-level semantic continuity and lexical boundary transitions separately. It achieves enhanced boundary prediction, reducing Pk by 0.7 points and Wd by 0.6 points on VHF under gold supervision, and by 14.8 points on VHF with induced boundaries, outperforming prior non-LLM approaches.
Dialogue topic segmentation is critical in many human-AI collaborative applications which requires identifying heterogeneous boundary cues, including lexical transitions near utterance edges and semantic discontinuities across utterances. Existing utterance models often dilute these local lexical signals. We propose CobSeg, a novel multi-branch architecture that separates coherence-level semantic continuity from lexical boundary transitions and recovers both through directional boundary prediction. CobSeg further uses boundary informativeness weighting to emphasize high-utility utterance positions, and incorporates a corpus-derived topic coherence cue with learned combination weights. While CobSeg is evaluated as a compact trainable segmenter under supervised gold-boundary training and a pseudo-label setting with automatically induced boundaries, it performs enhanced boundary prediction without LLM calls during inference. Across five benchmarks, it improves $P_k$ and $W_d$ particularly when local lexical cues are prominent: under gold supervision, it reduces $P_k$ by 0.7 points and $W_d$ by 0.6 points on VHF, and reaches $P_k$ of 1.0 on DialSeg711; with induced boundaries, it reduces $P_k$ by 14.8 points on VHF, by 1.5 points on DialSeg711, and by 1.1 points on TIAGE, outperforming prior non-LLM approaches.